Add gwp(1)

5 files changed, 86 insertions, 2 deletions

diff --git a/.gitignore b/.gitignore
index bd49f1fb..7bd19d59 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,5 +1,6 @@
 bin/csmw
 bin/ddup
+bin/gwp
 bin/han
 bin/mean
 bin/med
diff --git a/Makefile b/Makefile
index 1131af09..3f48b8d9 100644
--- a/Makefile
+++ b/Makefile
@@ -68,6 +68,7 @@ SENDMAIL := msmtp
 
 all : bin/csmw \
 	bin/ddup \
+	bin/gwp \
 	bin/han \
 	bin/mean \
 	bin/med \
@@ -87,6 +88,7 @@ clean distclean :
 	rm -f \
 		bin/csmw \
 		bin/ddup \
+		bin/gwp \
 		bin/han \
 		bin/mean \
 		bin/med \
@@ -185,8 +187,8 @@ install-bash-completion : install-bash
 	install -pm 0644 -- bash/bash_completion "$(HOME)"/.config/bash_completion
 	install -pm 0644 -- bash/bash_completion.d/* "$(HOME)"/.bash_completion.d
 
-install-bin : bin/csmw bin/ddup bin/han bin/mean bin/med bin/mftl bin/mode \
-	bin/rfct bin/rndi bin/sd2u bin/slsf bin/su2d bin/tot bin/unf \
+install-bin : bin/csmw bin/ddup bin/gwp bin/han bin/mean bin/med bin/mftl \
+	bin/mode bin/rfct bin/rndi bin/sd2u bin/slsf bin/su2d bin/tot bin/unf \
 	install-bin-man
 	install -m 0755 -d -- "$(HOME)"/.local/bin
 	for name in bin/* ; do \
diff --git a/README.markdown b/README.markdown
index fc248800..5eae69ac 100644
--- a/README.markdown
+++ b/README.markdown
@@ -444,6 +444,7 @@ Installed by the `install-bin` target:
 *   `grc(1df)` quietly tests whether the given directory appears to be a Git
     repository with pending changes.
 *   `gscr(1df)` scrubs Git repositories.
+*   `gwp(1df)` searches for alphanumeric words in a similar way to `grep(1)`.
 *   `han(1df)` provides a `keywordprg` for Vim's Bash script filetype that will
     look for `help` topics. You could use it from the shell too.
 *   `igex(1df)` wraps around a command to allow you to ignore error conditions
diff --git a/bin/gwp.awk b/bin/gwp.awk
new file mode 100644
index 00000000..db2764ec
--- /dev/null
+++ b/bin/gwp.awk
@@ -0,0 +1,57 @@
+#!/usr/bin/awk -f
+# Search for alphanumeric words in a file
+BEGIN {
+    
+    # Name self
+    self = "gwp"
+
+    # Words are separated by any non-alphanumeric character
+    FS = "[^[:alnum:]]"
+
+    # First argument is the word required; push its case downward so we can
+    # match case-insensitively
+    word = tolower(ARGV[1])
+
+    # Blank the first argument so Awk doesn't try to read data from it as a file
+    ARGV[1] = ""
+
+    # Bail out if we don't have a suitable word
+    if (!word)
+        fail("Need a single non-null alphanumeric string as a search word")
+    if (word ~ FS)
+        fail("Word contains non-alphanumeric characters; use grep(1)")
+}
+
+# Bailout function
+function fail(str) {
+    printf "%s: %s\n", self, str > "/dev/stderr"
+    exit(1)
+}
+
+# If there's more than one filename, precede the print of the current line with
+# a filename, a colon, and a space, much like grep(1) does; otherwise just
+# print it
+function fnpr() {
+    if (ARGC > 3)
+        print FILENAME ":" OFS $0
+    else
+        print
+}
+
+# Iterate through the words on this line and if any of them match our word,
+# print the line, and flag that we've found at least one word; once a single
+# instance of the word is found, just print and continue on to the next line
+{
+    for (i = 1; i <= NF; i++) {
+        if (tolower($i) == word) {
+            found = 1
+            fnpr()
+            break
+        }
+    }
+}
+
+# Exit zero if we found at least one match, non-zero otherwise
+END {
+    exit(!found)
+}
diff --git a/man/man1/gwp.1df b/man/man1/gwp.1df
new file mode 100644
index 00000000..c84cc12c
--- /dev/null
+++ b/man/man1/gwp.1df
@@ -0,0 +1,23 @@
+.TH GWP 1df "December 2016" "Manual page for gwp"
+.SH NAME
+.B gwp
+\- wordwise (alphanumeric) grep(1)
+.SH SYNOPSIS
+.B gwp WORD [FILE...]
+.br
+.SH DESCRIPTION
+.B gwp
+searches for complete alphanumeric words (not regular expressions) in the input and
+prints the line if found. This means you can search for "test" and it won't
+print lines just because they contain "latest". It's good for searching prose
+or poetry rather than code.
+.P
+This is intended as a workaround for the absence of a portable implementation
+of "word boundaries" in POSIX. Instead, this awk(1) script breaks each line
+down into alphanumeric words and tests each one for case-insensitive equality.
+.P
+It does not emulate all of grep(1)'s features by any means, but does include
+the feature of prefixing a matched line with the filename if more than one
+filename was searched.
+.SH AUTHOR
+Tom Ryder <tom@sanctum.geek.nz>